GAUSS Time Series#

A comprehensive time series analysis package for GAUSS, covering ARIMA/SARIMA, VAR/BVAR with stochastic volatility, structural identification, forecasting, and forecast evaluation.

Description#

GAUSS Time Series consolidates TSMT, SSLIB, and FANPAC into a single product. It provides:

Univariate models: ARIMA, SARIMA, ARIMAX with automatic order selection
Vector autoregression: OLS VAR, Bayesian VAR (Minnesota prior), BVAR with stochastic volatility
Structural identification: Cholesky IRF, generalized IRF, sign-restricted SVAR
Forecasting: Point, density, and conditional (scenario) forecasts
Model comparison: Marginal likelihood, Diebold-Mariano test, Model Confidence Set
Diagnostics: MCMC convergence (R-hat, ESS), forecast calibration (PIT)

Installation#

Please contact us for pricing and installation information.

Requires GAUSS v26 or higher.

Usage:

library timeseries;

Commands#

ARIMA / Univariate#

`arimaFit()`	Fit ARIMA, SARIMA, or ARIMAX models with automatic or fixed order selection.
`arimaForecast()`	Generate h-step-ahead forecasts with prediction intervals.
`arimaControlCreate()`	Create control structure with default settings.
`arimaResults()`	Reprint estimation summary table.
`arimaCoefTable()`	Return coefficient table as dataframe.

VAR Estimation#

`varFit()`	Fit VAR(p) by OLS with stability diagnostics.
`bvarFit()`	Fit Bayesian VAR with conjugate Minnesota or flat prior.
`bvarSvFit()`	Fit BVAR with stochastic volatility and optional SSVS variable selection.
`varLagSelect()`	Select lag order by AIC, BIC, or Hannan-Quinn.
`bvarHyperopt()`	Optimize Minnesota hyperparameters via marginal likelihood (GLP 2015).

Forecasting#

`varForecast()`	Point forecasts with confidence intervals from VAR.
`bvarForecast()`	Posterior predictive forecasts with credible bands.
`bvarSvForecast()`	Density forecasts from SV-BVAR with time-varying volatility.
`condForecast()`	Conditional (scenario) forecasts with hard constraints.

Impulse Responses & Structural Analysis#

`irfCompute()`	Orthogonalized (Cholesky) impulse response functions.
`irfSvCompute()`	Posterior IRF bands from SV-BVAR draws.
`girfCompute()`	Generalized IRF (Pesaran & Shin 1998), ordering-invariant.
`fevdCompute()`	Forecast error variance decomposition.
`hdCompute()`	Historical decomposition into structural shock contributions.
`irfPlotData()`	Reshape IRF results into plot-ready dataframe.

SVAR Identification#

`svarIdentify()`	Find a sign-restricted structural rotation.
`svarIrf()`	Posterior sign-restricted IRF, cumulative IRF, and FEVD bands.

Diagnostics#

`varDiagnose()`	MCMC convergence diagnostics (R-hat, ESS, acceptance rates).
`varDiagnoseMulti()`	Multi-chain convergence diagnostics.
`varDiagnosePrint()`	Reprint diagnostics summary.
`grangerTest()`	Granger causality F-test.

Forecast Evaluation#

`fcScore()`	Compute scoring rules (RMSE, MASE, sMAPE).
`dmTest()`	Diebold-Mariano test for equal predictive ability.
`cwTest()`	Clark-West test for nested model comparison.
`mcsTest()`	Model Confidence Set (Hansen, Lunde & Nason 2011).
`pitTest()`	PIT calibration tests (KS, chi-squared, Berkowitz).
`pitHistogram()`	PIT histogram bin counts.

Utilities#

`varCompanion()`	Extract companion matrix, eigenvalues, and stability indicator.
`varCoefTable()`	Return coefficient table as dataframe.
`varResults()`	Reprint estimation summary for any result type.
`stlDecompose()`	Seasonal-Trend decomposition via LOESS (STL).
`fcMetrics()`	Compute RMSE, MASE, and sMAPE.

Control Structure Creators#

`varControlCreate()`	Create `varControl` with defaults.
`bvarControlCreate()`	Create `bvarControl` with defaults.
`bvarSvControlCreate()`	Create `bvarSvControl` with defaults.
`svForecastControlCreate()`	Create `svForecastControl` with defaults.
`svarControlCreate()`	Create `svarControl` with defaults.